Abstract
We present spatial light interference microscopy (SLIM) as a new optical microscopy technique, capable of measuring nanoscale structures and dynamics in live cells via interferometry. SLIM combines two classic ideas in light imaging: Zernike’s phase contrast microscopy, which renders high contrast intensity images of transparent specimens, and Gabor’s holography, where the phase information from the object is recorded. Thus, SLIM reveals the intrinsic contrast of cell structures and, in addition, renders quantitative optical path-length maps across the sample. The resulting topographic accuracy is comparable to that of atomic force microscopy, while the acquisition speed is 1,000 times higher. We illustrate the novel insight into cell dynamics via SLIM by experiments on primary cell cultures from the rat brain. SLIM is implemented as an add-on module to an existing phase contrast microscope, which may prove instrumental in impacting the light microscopy field at a large scale.
OCIS codes: (180.3170) Interference microscopy, (180.0180) Microscopy, (999.9999) Quantitative phase imaging
1. Introduction
Most living cells do not absorb or scatter light significantly, i.e. they are essentially transparent, or phase objects. Phase contrast microscopy proposed by Zernike represented a major advance in intrinsic contrast imaging, as it revealed inner details of transparent structures without staining or tagging [1]. While phase contrast is sensitive to minute optical path-length changes in the cell, down to the nanoscale, the information retrieved is only qualitative. Quantifying cell-induced shifts in the optical path-lengths permits nanometer scale measurements of structures and motions in a non-contact, non-invasive manner [2]. Thus, quantitative phase imaging (QPI) has recently become an active field of study and various experimental approaches have been proposed [3–11]. Advances in phase-sensitive measurements enabled optical tomography of transparent structures, following reconstruction algorithms borrowed from X-ray computed imaging, in which scattering and diffraction effects are assumed to be negligible [12–16]. Further, QPI-based projection tomography has been applied to live cells [17–19].
Despite these significant technological advances, the range of QPI applications in biology has been largely limited to red blood cell imaging [20–22] or assessment of global cell parameters such as dry mass [4,23], average refractive index [24], and statistical parameters of tissue slices [25]. This limitation is due to two main reasons, as follows. First, because of speckle generated by the high temporal coherence of the light used (typically lasers), the contrast in QPI images has never matched that exhibited in white light techniques such as phase contrast and Nomarski. Second, the experimental setups tend to be rather complex, of high maintenance, which limits their in-depth biological applicability.
Here, we present SLIM as a novel, highly sensitive QPI method, which promises to enable unprecedented structure and dynamics studies in biology and beyond. SLIM combines Zernike’s phase contrast method by revealing the intrinsic contrast of transparent samples [1], with Gabor’s holography [26] by rendering quantitative phase maps across the sample. Because of the extremely short coherence length of this illumination light, approximately 1.2 μm, SLIM provides speckle-free imaging with sub-nanometer spatial background noise. Further, the SLIM image is intrinsically registered with the other channels of the microscope, including fluorescence, which enables powerful multimodal investigations.
2. SLIM Setup
A schematic of the instrument setup is depicted in Fig. 1a . SLIM was developed by producing additional spatial modulation to the image field outputted by a commercial phase contrast microscope. Specifically, in addition to the π/2 shift introduced in phase contrast between the scattered and unscattered light from the sample [1], we generated further phase shifts, by increments of π/2, and recorded additional images for each phase map. Thus, the objective exit pupil, containing the phase shifting ring, is imaged via lens L1 onto the surface of a reflective liquid crystal phase modulator (LCPM, Boulder Nonlinear). The active pattern on the LCPM is calculated to precisely match the size and position of the phase ring image, such that additional phase delay can be precisely controlled between the scattered and unscattered components of the image field. In this setup, 4 images corresponding to each phase shift are recorded (Fig. 1b), to produce a quantitative phase image that is uniquely determined (see LCPM calibration Section below). Figure 1c depicts the quantitative phase image associated with a cultured hippocampal neuron, which is proportional to
(1) |
In Eq. (1), n-n0 is the local refractive index contrast between the cell and the surrounding culture medium, h the local thickness of the cell, and λ the central wavelength of the illumination light. SLIM provides the local phase shift ϕ with great accuracy, which in turn allows for detecting local changes in thickness h at a scale much smaller than the wavelength of light. The typical irradiance at the sample plane is ~1 nW/ μm2. The exposure time was 10-50 ms, for all the images presented in the manuscript. This level of exposure is 6-7 orders of magnitude below that of typical confocal microscopy [27], which allows for noninvasive live-cell imaging over extended periods of time (many hours or days).
3. Liquid crystal phase modulation
Using a spectrometer (USB 2000, Ocean Optics, USA), we measured the optical spectrum of the illuminating white light at the CCD plane. This spectrum is shown in Fig. 2a . From these data, we retrieved the temporal autocorrelation function of the light via a Fourier transform (Fig. 2b). The spectrum provided by the spectrometer is sampled in wavelength. In order to obtain the spectrum vs. frequency, which then can be Fourier transformed, we performed resampling of the data. In a medium of refractive index n = 1.33 (i.e. water), the coherence length defined at full width half maximum is . This coherence length is at least an order of magnitude shorter than that of other light sources such as broadband lasers, light emitting diodes and superluminescent diodes. However, within this coherence length there are still several full cycle modulations. Thus, the envelop varies slowly over one period near the central peak, which enables the application of the phase shifting procedure described below.
The LCPM (XY Phase Series Model P512 –635, Boulder Nonlinear Systems, Inc, USA) was calibrated to decide the relationship between pixel grey values fed via a VGA signal to the liquid crystal array and the final phase delay introduced to the unscattered field. The LCPM was placed between two polarizers and its intensity transmission was recorded as follows. We first changed the polarizer and analyzer by 45° so that SLM will work in “amplitude modulation” mode. Then we scanned through the grayscale value from 128 to 255 (i.e. 8 bits). The modulation from pixel value 0 to 127 and from 128 to 255 is symmetric. Thus we only need to scan half of the pixel values. The intensity transmitted through the LCPM vs. the grey value is shown in Fig. 2c. From the amplitude response of the modulator we obtained its phase response via a Hilbert transform, as shown in Fig. 2d.
For spatial coherent imaging system, after passing the specimen, a portion of the light remains unscattered and forms a uniform background of the image; the other portion is scattered and contains the fine structure information of the specimen. If we denote the unscattered light as and the scattered light field as , the intensity at CCD is
(2) |
where is the phase difference between and , and φ is the additional phase modulation introduced by LCPM, as shown in Fig. 2. Note that for φ = π/2, traditional phase contrast is obtained.
The quantity of interest is retrieved as
(3) |
If we define , then the phase associated with the image field can be determined [11,28]:
(4) |
Note that ϕ defines the phase delay associated with the autocorrelation function depicted in Fig. 2b.
4. Nanoscale topography with SLIM
In order to assess the spatial accuracy of SLIM, we imaged an amorphous carbon film deposited on glass and compared the topography measurements against atomic force microscopy (AFM). The topography measurements by SLIM and AFM, respectively, are summarized in Fig. 3a and 3b. The two types of measurement agree within a fraction of a nanometer. Note that both SLIM and AFM are characterized by much smaller errors than suggested by the widths of the histogram modes, as these widths also reflect irregularities in the surface topography due to the fabrication process itself. Of course, AFM is not limited by diffraction in the transverse direction. Unlike AFM, SLIM is non-contact, parallel, and faster by more than 3 orders of magnitude. Thus, SLIM can optically measure an area of 75 × 100 μm2 in 0.5 s compared to a 10 × 10 μm2 field of view measured by AFM in 21 minutes.
As further comparison, we evaluated background images (i.e., no sample) from SLIM and diffraction phase microscopy (DPM) [9], an established laser-based technique that was interfaced with the same microscope (Fig. 3d and 3e). Due to the lack of speckle effects granted by its broad spectral illumination, SLIM’s spatial uniformity and accuracy for structural measurements is substantially better than DPM’s. To quantify the spatio-temporal phase sensitivity, we imaged the SLIM background repeatedly to obtain a 256-frame stack. Figure 3f shows the spatial and temporal histograms associated with the optical path-length shifts across a 10 × 10 μm2 field of view and over the entire stack, respectively. These noise levels, 0.3 nm and 0.03 nm, represent the limit in optical path-length sensitivity across the image and between frames, respectively. The diminished effects of speckles allowed quantitative phase imaging to reveal single atomic layers of carbon, for the first time [29]. Several error sources can potentially be diminished further: residual mechanical vibrations in the system that are not “common path”, minute fluctuations in the intensity and spectrum of the thermal light source, digitization noise from the CCD camera (12 bit in our case), and the stability (repeatability) of the liquid crystal modulator (8 bit).
The LCPM maximum refresh rate is 60 Hz, in principle allowing for 15 SLIM images per second but throughout the manuscript we report imaging at 2.6 frames/ s, as our camera has a maximum acquisition rate of 11 frames/s at full resolution. Acquisition speed could be increased to video rate by employing a faster phase modulator and camera.
5. Multimodal SLIM-fluorescence imaging
A distinct feature of SLIM is that the quantitative phase image is overlaid with the other imaging channels of the microscope, such as epi-fluorescence, differential interference contrast, and phase contrast. Simultaneous fluorescence imaging complements SLIM’s structural information with the ability to study cellular constituents with molecular specificity. In Fig. 4a and 4b, we show multimodal SLIM-fluorescence imaging of primary hippocampal neurons cultured for 19 days in vitro (DIV). Nucleus (blue) and dendrites (green) are stained by 4',6-diamidino-2-phenylindole (DAPI) and the somatodendritic marker microtubule-associated protein 2 (MAP2), respectively. The red line in the inset of Fig. 4a indicates the axon, as evidenced by the absence of MAP2 staining.
In order to quantify the structures observed by SLIM, we traced individual neurites using NeuronJ (Fig. 4c). Each trace shows the optical path-length fluctuations along each different neurite. The standard deviation (σ) of the path-length fluctuation is, on average, twice as large for dendrites as it is for the axon, i.e., 49 nm vs. 25 nm. This result suggests that subtle inhomogeneities are associated with the synaptic structures and these inhomogeneities can be revealed by SLIM as path-length changes. By 3 weeks in dispersed culture, the majority of dendritic spines generally mature to form presynaptic boutons on the dendritic shafts. After 33 DIV, we observe synapsin and MAP2 labeling of putative synaptic elaborations on a mature hippocampal neuron (Fig. 4d).
From the quantitative phase measured, other representations of the information can be obtained numerically. Thus, the spatial gradient of the SLIM image simulates DIC microscopy in Fig. 5c ( Media 1 (9.3MB, MOV) ). Further, we show that the Laplacian of the image, a second order derivative operation, is even more powerful than DIC in revealing fine structures within the cell, as it does not contain significant shadow artifacts in Fig. 5d ( Media 1 (9.3MB, MOV) ).
6. Cell dynamics measurements
Due to the extremely low level of spatial noise (0.3 nm) and temporal stability (0.03 nm), SLIM is able to image sub-nanometer dynamics in live cells for periods ranging from seconds to days. For example, we obtained 397 two-dimensional SLIM images of a mixed glial-microglial cell culture over 13 minutes (Fig. 6 ). When we compare live-cell measurements made by phase contrast vs. SLIM imaging, significant differences are apparent. The cell size is clearly overestimated by phase contrast due to the well known halo artifact which makes the borders of the cell appear bright (Fig. 6b). Also, the conventional phase contrast images extracted from these data are unable to provide quantitative information about dynamic changes in optical path-length (Fig. 6b and 6c). Importantly, due to the contrast change of the halo with the phase shift (Fig. 1b), SLIM is able to minimize this effect when combining the four phase ring frames (Fig. 6d).
Path-length changes due to both membrane displacements and local refractive index changes caused by cytoskeleton dynamics and particle transport at two arbitrary points on the cell reveal an interesting, periodic behaviour (Fig. 6d, 6f). At different sites on the cell, the rhythmic motions have different periods, which may indicate different rates of metabolic or phagocytic activity. This periodicity can be observed in coordinated cell behaviour as the cell extends broad, dynamic filopodial ruffles under, and above, the neighbouring glial cells in Fig. 6 ( Media 2 (10MB, MOV) ).
Because of the extremely low noise level of SLIM, the probability distribution of path-length displacements between two successive frames was retrieved with a dynamic range of over 5 orders of magnitude (Fig. 6g). This distribution can be fitted very well with a Gaussian function up to path-length displacements Δs = 10 nm, at which point the curve crosses over to an exponential decay. The normal distribution suggests that these fluctuations are the result of numerous uncorrelated processes governed by equilibrium. On the other hand, exponential distributions are indicative of deterministic motions, mediated by metabolic activity.
SLIM studies of cell dynamics should reveal previously unknown information regarding membrane motions, cytoskeleton mechanics, and particle transport within the cell. Figure 7 ( Media 3 (9.7MB, MOV) ) shows SLIM imaging of neuron processes, which suggests that SLIM may offer a window into studying the dynamic processes associated with the formation and transition of collateral filopodia into spines, and the dynamics of plasticity-related changes in spine structure [30]. Note that SLIM can be used to image cellular dynamics, including bidirectional neurite transport and actin dynamics over extended periods of time without loss in performance.
7. Summary and outlook
In sum, our results demonstrate that rich and quantitative information can be captured from biological structures using SLIM without physical contact or staining. The main features associated with SLIM are as follows. SLIM provides speckle-free images, which allows for spatially sensitive optical path-length measurement (0.3 nm). It uses common path interferometry, which enables temporally sensitive optical path-length measurement (0.03nm). Our initial results demonstrate that rich, previously unobservable information can be captured in the spatially-resolved cell phase shift map as demonstrated by neuron imaging. The co-registration with the fluorescence channels of the microscope represents a significant feature, which enables in-depth biological studies. Finally, the broadband illumination field offers opportunities for spectroscopy, as recently demonstrated in a different configuration [31]. The authors are optimistic about SLIM having a major impact in the field of light microscopy, as existing phase contrast microscopes can be converted easily (via an add-on module) to provide quantitative and nanoscale information.
Acknowledgments
This study was supported by the National Science Foundation ( CAREER 08-46660 to GP), the Grainger Foundation (to GP), and the National Institute of Mental Health ( R21 MH085220 to MUG). LM was supported by the National Institute of Child Health and Human Development Developmental Psychobiology and Neurobiology Training Grant ( HD007333). ZW and GP acknowledge visualization tools by Ethan Berl and stimulating discussions with Dan Marks and Scott Carney. Related information can be found at http://light.ece.uiuc.edu/.
References and links
- 1.Zernike F., “How I discovered phase contrast,” Science 121(3141), 345–349 (1955). 10.1126/science.121.3141.345 [DOI] [PubMed] [Google Scholar]
- 2.G. Popescu, “Quantitative phase imaging of nanoscale cell structure and dynamics,” in Methods in Cell Biology, P. J. Bhanu, ed. (Elsevier, 2008), p. 87.3128434 [DOI] [PubMed] [Google Scholar]
- 3.Paganin D., Nugent K. A., “Noninterferometric phase imaging with partially coherent light,” Phys. Rev. Lett. 80(12), 2586–2589 (1998). 10.1103/PhysRevLett.80.2586 [DOI] [Google Scholar]
- 4.Zicha D., Dunn G. A., “An Image-Processing System For Cell Behavior Studies In Subconfluent Cultures,” J. Microsc. 179, 11–21 (1995). 10.1111/j.1365-2818.1995.tb03609.x [DOI] [Google Scholar]
- 5.Yang C. H., Wax A., Dasari R. R., Feld M. S., “Phase-dispersion optical tomography,” Opt. Lett. 26(10), 686–688 (2001). 10.1364/OL.26.000686 [DOI] [PubMed] [Google Scholar]
- 6.Choma M. A., Ellerbee A. K., Yang C. H., Creazzo T. L., Izatt J. A., “Spectral-domain phase microscopy,” Opt. Lett. 30(10), 1162–1164 (2005). 10.1364/OL.30.001162 [DOI] [PubMed] [Google Scholar]
- 7.Joo C., Akkin T., Cense B., Park B. H., de Boer J. F., “Spectral-domain optical coherence phase microscopy for quantitative phase-contrast imaging,” Opt. Lett. 30(16), 2131–2133 (2005). 10.1364/OL.30.002131 [DOI] [PubMed] [Google Scholar]
- 8.Rockward W. S., Thomas A. L., Zhao B., Dimarzio C. A., “Quantitative phase measurements using optical quadrature microscopy,” Appl. Opt. 47(10), 1684–1696 (2008). 10.1364/AO.47.001684 [DOI] [PubMed] [Google Scholar]
- 9.Popescu G., Ikeda T., Dasari R. R., Feld M. S., “Diffraction phase microscopy for quantifying cell structure and dynamics,” Opt. Lett. 31(6), 775–777 (2006). 10.1364/OL.31.000775 [DOI] [PubMed] [Google Scholar]
- 10.Ikeda T., Popescu G., Dasari R. R., Feld M. S., “Hilbert phase microscopy for investigating fast dynamics in transparent systems,” Opt. Lett. 30(10), 1165–1167 (2005). 10.1364/OL.30.001165 [DOI] [PubMed] [Google Scholar]
- 11.Popescu G., Deflores L. P., Vaughan J. C., Badizadegan K., Iwai H., Dasari R. R., Feld M. S., “Fourier phase microscopy for investigation of biological structures and dynamics,” Opt. Lett. 29(21), 2503–2505 (2004). 10.1364/OL.29.002503 [DOI] [PubMed] [Google Scholar]
- 12.Chen B. Q., Stamnes J. J., “Validity of diffraction tomography based on the first born and the first rytov approximations,” Appl. Opt. 37(14), 2996–3006 (1998). 10.1364/AO.37.002996 [DOI] [PubMed] [Google Scholar]
- 13.Gbur G., Wolf E., “Relation between computed tomography and diffraction tomography,” J. Opt. Soc. Am. A 18(9), 2132–2137 (2001). 10.1364/JOSAA.18.002132 [DOI] [PubMed] [Google Scholar]
- 14.Carney P. S., Wolf E., Agarwal G. S., “Diffraction tomography using power extinction measurements,” J. Opt. Soc. Am. A 16(11), 2643–2648 (1999). 10.1364/JOSAA.16.002643 [DOI] [Google Scholar]
- 15.Lauer V., “New approach to optical diffraction tomography yielding a vector equation of diffraction tomography and a novel tomographic microscope,” J. Microsc. 205(2), 165–176 (2002). 10.1046/j.0022-2720.2001.00980.x [DOI] [PubMed] [Google Scholar]
- 16.Zysk A. M., Reynolds J. J., Marks D. L., Carney P. S., Boppart S. A., “Projected index computed tomography,” Opt. Lett. 28(9), 701–703 (2003). 10.1364/OL.28.000701 [DOI] [PubMed] [Google Scholar]
- 17.Charrière F., Pavillon N., Colomb T., Depeursinge C., Heger T. J., Mitchell E. A. D., Marquet P., Rappaz B., “Living specimen tomography by digital holographic microscopy: morphometry of testate amoeba,” Opt. Express 14(16), 7005–7013 (2006). 10.1364/OE.14.007005 [DOI] [PubMed] [Google Scholar]
- 18.Charrière F., Marian A., Montfort F., Kuehn J., Colomb T., Cuche E., Marquet P., Depeursinge C., “Cell refractive index tomography by digital holographic microscopy,” Opt. Lett. 31(2), 178–180 (2006). 10.1364/OL.31.000178 [DOI] [PubMed] [Google Scholar]
- 19.Choi W., Fang-Yen C., Badizadegan K., Oh S., Lue N., Dasari R. R., Feld M. S., “Tomographic phase microscopy,” Nat. Methods 4(9), 717–719 (2007). 10.1038/nmeth1078 [DOI] [PubMed] [Google Scholar]
- 20.Popescu G., Ikeda T., Goda K., Best-Popescu C. A., Laposata M., Manley S., Dasari R. R., Badizadegan K., Feld M. S., “Optical measurement of cell membrane tension,” Phys. Rev. Lett. 97(21), 218101 (2006). 10.1103/PhysRevLett.97.218101 [DOI] [PubMed] [Google Scholar]
- 21.Park Y. K., Diez-Silva M., Popescu G., Lykotrafitis G., Choi W., Feld M. S., Suresh S., “Refractive index maps and membrane dynamics of human red blood cells parasitized by Plasmodium falciparum,” Proc. Natl. Acad. Sci. U.S.A. 105(37), 13730–13735 (2008). 10.1073/pnas.0806100105 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Y. K. Park, C. A. Best, K. Badizadegan, R. R. Dasari, M. S. Feld, T. Kuriabova, M. L. Henle, A. J. Levine, and G. Popescu, “Measurement of red blood cell mechanics during morphological changes,” Proc. Nat. Acad. Sci. (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Popescu G., Park Y., Lue N., Best-Popescu C., Deflores L., Dasari R. R., Feld M. S., Badizadegan K., “Optical imaging of cell mass and growth dynamics,” Am. J. Physiol. Cell Physiol. 295(2), C538–C544 (2008). 10.1152/ajpcell.00121.2008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Lue N., Popescu G., Ikeda T., Dasari R. R., Badizadegan K., Feld M. S., “Live cell refractometry using microfluidic devices,” Opt. Lett. 31(18), 2759–2761 (2006). 10.1364/OL.31.002759 [DOI] [PubMed] [Google Scholar]
- 25.Ding H., Nguyen F., Boppart S. A., Popescu G., “Optical properties of tissues quantified by Fourier-transform light scattering,” Opt. Lett. 34(9), 1372–1374 (2009). 10.1364/OL.34.001372 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Gabor D., “A new microscopic principle,” Nature 161(4098), 777–778 (1948). 10.1038/161777a0 [DOI] [PubMed] [Google Scholar]
- 27.J. B. Pawley, Handbook of biological confocal microscopy (Springer, New York, 2006). [Google Scholar]
- 28.Wang Z., Popescu G., “Quantitative phase imaging with broadband fields,” Appl. Phys. Lett. 96(5), 051117 (2010). 10.1063/1.3304787 [DOI] [Google Scholar]
- 29.Wang Z., Chun I. S., Li X. L., Ong Z. Y., Pop E., Millet L., Gillette M., Popescu G., “Topography and refractometry of nanostructures using spatial light interference microscopy,” Opt. Lett. 35(2), 208–210 (2010). 10.1364/OL.35.000208 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Waites C. L., Craig A. M., Garner C. C., “Mechanisms of vertebrate synaptogenesis,” Annu. Rev. Neurosci. 28(1), 251–274 (2005). 10.1146/annurev.neuro.27.070203.144336 [DOI] [PubMed] [Google Scholar]
- 31.Ding H. F., Popescu G., “Diffraction phase contrast microscopy,” Opt. Express 18(2), 1569–1575 (2010). 10.1364/OE.18.001569 [DOI] [PubMed] [Google Scholar]